EFFECTS OF URBAN GREENERY ON HEALTH.

A STUDY FROM REMOTE SENSING

Blanca Arellano 1, Josep Roca 2 *
1 Technical University of Catalonia, blanca.arellano@upc.edu
2 Technical University of Catalonia, josep.roca@upc.edu

ICWG III/1: Thematic Information Extraction

KEY WORDS: Urban Greenery, Nighttime Urban Heat Island, LST, Health, Landsat 8, Sentinel 2, Metropolitan Area of Barcelona

ABSTRACT:

Global warming is causing increasing Heat Waves that affect human health. High temperatures markedly increase morbidity and
mortality. Urban Heat Islands increase the effects of Heat Waves and are a serious inconvenience to human health and comfort. Cities
can substantially increase local temperatures and reduce temperature drop at night. During the night, the greater thermal inertia of the
central areas reduces their cooling capacity. On the other hand, it is important to highlight that urban vegetation plays a key role in
adapting cities to Global Warming and Urban Heat Island. Green areas have lower temperatures than the rest of land uses and generate
a cooling effect that spreads to their surroundings creating a "cool island" effect. The main objective of this paper is to establish the
nocturnal land surface temperature and land surface air temperature of Barcelona Metropolitan Area (35 municipalities, 636 km2, 3.3
million inhabitants) in an episode of a nocturnal heatwave and to estimate its possible impact on health and mortality. Subsequently,
nighttime temperatures are analysed in this extreme heat context to determine their spatial distribution and detect the urban landscapes
that are most vulnerable to extreme night heat. Modelling of land surface temperature must reveal the elements that determine night
Urban Heat Island and consequently identify actions that can be implemented at urban planning level to refresh the environment during
the night and thus increase the resilience of the most vulnerable landscapes and improve residents’ health. This paper studies the effect
of urban greenery and green infrastructures on Nighttime Urban Heat Island and propose climate adaptation measures and design for
urban green areas to decrease high temperature in a Heat Wave context, which contributes to reducing the serious negative impacts on
people's health.

1. INTRODUCTION (Eurostat, 2022). Climate change can therefore affect human

Global warming (GW) is causing increasing heatwaves (HW)
that affect human health (IPCC, 2012). Climate change (CC)
increases the risk of extreme temperatures and weather events,
such as heatwaves (HW). Summer heatwaves have increased in
frequency and duration in most of Europe while extreme cold
temperatures and cold waves are less frequent. Trends show that
in the future with a warmer climate and increased mean
temperatures, heatwaves will become more intense, longer
lasting, and/or more frequent (Meehl and Tebaldi, 2004). As
IPCC (2021) says: “It is virtually certain that the frequency and
intensity of hot extremes and the intensity and duration of
heatwaves have increased since 1950 and will further increase in
the future even if global warming is stabilized at 1.5ºC.”. Hence,
it is highly likely that CC causes an increase in HWs, which
negatively affect human health.
The effects of heatwaves on the population have been described
by numerous authors who have established clear relationships
between high temperatures and morbidity and mortality,
especially in respiratory and cardiovascular diseases (Royé,
2017). The human organism and the atmosphere are in a
constantly interacting physical and chemical equilibrium. All
humans are forced to react to atmospheric elements to guarantee
the correct, optimum functioning of their organs. The thermal
environment plays a significant role in public health (Parsons,
2014). Sociodemographic and urban landscape characteristics are
associated with mortality risk during heatwaves (Gasparrini et al,
2015). In much of the developed world, societies are aging and
hence can be more vulnerable to climate extremes such as
heatwaves. For example, Europe currently has an aging
population, with a higher population density and lower birth rate
health by changing the severity or frequency of health problems
that are already affected by climate factors. It can create
unprecedented or unanticipated health problems or health threats
in places where they have not previously occurred.
Although the climate of cities depends fundamentally on
mesoscale factors, local and microscale factors can modify
regional climate at urban scale. These factors include the
characteristics of the urban structure, the topography and surface
of roofs, vegetation, and anthropogenic heat generated by urban
metabolism. There are significant differences in the climate of
urban areas compared to those of a rural nature. The urban heat
island effect (UHI) describes the influence of urban surfaces on
the temperature patterns of urban areas as opposed to surrounding
areas (Oke, 1967). Cities accumulate heat in urban land covers
and built infrastructures, representing true islands of heat in
relation to the rural (less artificialized) environment. The densest
urban spaces and industrial and commercial areas are
characterized by accumulating more heat during the daytime. In
this context, urban heat islands increase the effects of HW.
Urban heat islands increase the effects of heatwaves, representing
a serious inconvenience to human health and comfort (Tan et al,
2010). Cities can substantially increase local temperatures and
reduce temperature drop at night. During the night, the greater
thermal inertia of the central areas reduces their cooling capacity,
which represents a serious inconvenience in the case of “tropical”
(> 20° Celsius) and "torrid" nights (> 25° Celsius). In cases of
high nighttime temperatures, as a consequence of high daytime
temperatures, heat stress persists and is impossible to rest and
therefore sleep. The literature on urban climate has highlighted
the importance of the nighttime UHI. It is during the night that

* Corresponding author.
the effects of UHI become more apparent, due to the low cooling
capacity of urban construction materials and is during nighttime
that temperatures can cause higher health risks. However, the
study of nocturnal UHIs is still poorly developed, due to the
structural problems regarding the availability of land surface
temperature (LST) and land surface air temperature (LSAT) data
for night time. Traditional methods for obtaining nocturnal UHI
have been directed either to extrapolation of data from weather
stations, or obtaining air temperatures through urban transects. In
the first case, the lack of weather stations in urban landscapes
makes it extremely difficult to obtain data to extrapolate and
propose models at a detailed resolution scale. In the second case,
there is a manifest difficulty in obtaining data simultaneously and
significantly representative of urban and rural zones.
The use of satellite images has helped considerably to understand
UHI, especially in analyses of daytime land surface temperature
(LST). However, available nighttime sensors have a major
limitation: their low spatial resolution does not allow detailed
analysis of nighttime UHI. In general, few studies have focused
on studying nocturnal UHI, due to the lack of information on
LST. MODIS and Sentinel 3, for example, have a spatial
resolution of 1000 meters/pixel, which is generally insufficient
for the analysis of the spatial distribution of the UHI, which has
led to the need to downscale said spatial resolution (Mahour et
al, 2017; Guzinski et al, 2020). High resolution thermal images
(Landsat, for example) are limited to daytime information. In
general, few studies have focused on studying Nocturnal Urban
Heat Island [NUHI], due to the lack of information on nighttime
thermal imagery (Arellano & Roca, 2021). The study of the
nocturnal LST is of great importance in heat stress events derived
from Night Heat Waves.
Furthermore, the integration of LST (obtained from remote
sensing imagery) with LSAT (obtained from weather stations)
continues to be a pending challenge. The right estimation of the
temperature of the air at ≈ 2-m height above ground (LSAT) from
LST is possible but complex. The vertical lapse rate to be applied
is function of the surface energy balance, which varies in function
of the nature of the surface and of the instant of the day, as also
of advection, adiabatic processes, turbulence and latent heat
fluxes, all of them affected by cloud cover, water vapour content
and vegetation (Benali, 2012). At night, however, there is a
greater homogenization of advection and adiabatic processes,
which represents less complexity in the modeling of air
temperatures. The large differences between land surface and air
temperatures during the day tend to moderate at night, with the
LST and LSAT converging notably, especially in summer.
Air temperature estimated from satellite measurements would
solve the weather stations scarcity in wider regions, where the
geospatial interpolation methods cannot provide accurate
estimations. In this way, LSAT estimation becomes of crucial
importance to solve spatial gaps for a wide range of applications,
in such a way that it is accepted that TIR produces better LSAT
estimations than those obtained by interpolating ground-station
temperatures (Mendelsohn et al, 2007). There is significant
literature that has deepened on how to obtain the air temperature
from MODIS LST. For example, the construction of multiple
regression models with air temperature as the dependent variable
and daytime and nighttime LSTs, along with other variables such
as altitude, longitude, latitude, distance to the sea, NDVI or
albedo, as independent variables (Serra et al, 2020). However,
there are few studies that have managed to increase the resolution
of the air temperature models, due to the low resolution of the
data provided by MODIS or Sentinel 3 (Arellano et al, 2021).
This paper seeks to determine the nocturnal land surface and air
temperature in a period of maximum nighttime heat (summer
2015). So, this research aims to develop a nighttime land surface
air temperature (NLSAT) model by merging information from
Landsat, MODIS and information from meteorological stations
in order to study the contribution Night Urban Heat Island
(NUHI) in episodes of heat wave. Given the practical identity that
exists (at night) between air temperature (that is, that experienced
by humans) and the temperature of the earth's surface (obtained
by remote sensing), the study of the nocturnal LST is of great
importance in thermal stress events derived from Night Heat
Waves (NHW).
On the other hand, it is important to highlight that urban
vegetation plays a key role in adapting cities to climate change.
Greenery increases air humidity and due to the green canopy the
shaded areas. Such characteristics break the UHI continuum,
which is the rise in temperature that cities present, partly, because
of high absorption of direct sun-heat in artificial surfaces. Thus,
green areas have lower temperatures than the rest of land uses
and generate a cooling effect that spreads to their surroundings
creating a "cool island" effect (Bowler et al, 2010).
The cooling effect of parks is quantified by the extent limit,
which is the maximum distance reached by the cooling spread
outside boundaries of the park; and the intensity, which is the
difference in temperature between the park and a certain urban
space in its near surroundings (Spronken-Smith & Oke, 1998).
There is a consensus on the calculation of the cooling intensity,
but with slight differences on the spatial attribution of the
temperature to the park and the one that represents the urban
space. Previous investigations pointed that the cooling extent of
parks between 3 and 200 ha size, is in the 50 to 300 meters (m)
range, but larger parks go from 200 to 2000m. Whereas the
cooling intensity registers values between 1 to 4°C during day
and 2 to 5°C at night (Kuttler, 2012), with an average intensity
between 0.94 and 1.15°C (Bowler et al, 2010).
From this perspective, the research aims, through the use of
remote sensing, to quantify the influence of the physical
characteristics of urban parks and their surroundings to improve
the criteria of climate-sensitive urban design and planning and
reduce risks for the health of its inhabitants during extreme heat
waves. In this context, the paper studies the effect of urban
greenery and green infrastructures on night-time UHI in
Metropolitan Area of Barcelona (MAB, fig. 1), and propose
climate adaptation measures and design for urban green areas to
decrease high temperature in a Heat Wave context, which
contributes to reducing the serious negative impacts on people's
health.
Figure 1. Metropolitan Area of Barcelona (with municipalities)
 2. METHODOLOGY
The methodology used can be summarized as follows:
1. First of all, the heat waves experienced in the city of
Barcelona are studied, in order to determine the most
appropriate period to carry out the analysis of the LST and
the LSAT. There is no universal definition of a heatwave,
but extreme events associated with particularly hot,
sustained temperatures have been known to have a notable
impact on human mortality, regional economies and
ecosystems. In this paper, we use the concept of heatwave
applied by the Spanish Meteorological Agency (AEMET).
In this definition, a heatwave is considered an episode of at
least three consecutive days in which at least 10% of the
stations that are considered register maximums above the
95% percentile of the series of maximum daily temperatures
for the months of July and August from the period 1971 to
2000. However, this definition has a major limitation: it
refers only to maximum temperatures, not minimum ones.
As indicated, it is the high minimum temperatures that make
the most difference for health purposes. Maximum
temperatures can have serious consequences, especially on
heat stroke. However, the health effects are more
pronounced in the case of night heat, where the inability to
rest (especially in homes without air conditioning, as is
generally the case in in Spain) can cause significant
worsening of respiratory and cardiovascular diseases that
produce a high proportion of premature deaths. For this
reason, in this study we differentiate between heatwaves
during the day (DHW) and at night (NHW), paying special
attention to the latter.
2. The second step of the study consisted of determining the
nighttime LST of Barcelona in a heatwave. August 2015
was chosen since it had temperatures that were higher than
the average of the preceding years. There were various day
and night heatwaves. This date was also selected due to the
availability of night thermal images (bands 10 and 11) of
Landsat 8, which allow the night LST of Barcelona
Metropolitan Area to be obtained in detail. The day and
night images of Landsat 8 from 3 and 28 August 2015 were
used, complemented by the night image of 13 September
2015 (to achieve adequate exclusion of areas covered by
clouds).
3. The estimation of LST from band 10 of Landsat 8 was
carried out according to the abundant existing literature on
the matter. However, the estimation of nighttime land
surface temperature by Landsat thermal bands is not a trivial
question. The most commonly used methodology to
determine daytime LST is based on estimating the
emissivity of the land from its degree of vegetation: the
normalized difference vegetation index (NDVI) threshold
(Sobrino et al, 2008). But this method has significant
limitations at night. The NDVI overvalues vegetation when
it considers the canopy of trees. This overestimation is not a
serious problem during the day, when the shade of trees
limits the incident radiation on the ground. However, it is
critical at night. At night, the application of the standard
method for determining emissivity underestimates the land
surface radiation in areas with abundant tree vegetation. The
result is critical on the streets and in parks with a significant
degree of canopy. For this reason, this paper used the
methodology suggested by (Arellano & Roca, 2021) to
estimate the degree of vegetation and soil moisture. Based
on this estimation, the emissivity was determined and
consequently the nocturnal LST. In this paper, winter NDVI
was used instead of summer NDVI, as it better represents
the degree of vegetation at ground level. The tree canopy
plays a smaller role due to the deciduous nature of the trees
in the study area. At night, when solar radiation is non-
existent, long-wave radiation from the ground is not
significantly affected by the tree canopy.
4. Modeling the spatial distribution of the LST. The
methodology used in this work has been to develop a
"hybrid" OLS model, with geographical (longitude, latitude,
altitude, slope, orientation and distance to the sea) and
urban-territorial (such as NDVI, NDBI, albedo,
imperviousness, land covers, …) explanatory variables. The
MODIS LST is also included in the model. Said "hybrid"
model enables not only LST downscaling (from 1000
meters, MODIS, to 30 meters, Landsat), but also the
possible application to other moments in time to those of the
date of acquisition of the image.
5. On the other hand, directly modeling the existing
information on the land surface air temperature resulting
from the meteorological stations. The LSAT is obtained (for
the date of the Landsat 8 image) based on the methodology
developed by (Serra et al, 2020). The daily MOD11A1 LST
measured by MODIS Terra, including daytime (LSTd) and
nighttime (LSTn) with 1 km2 spatial resolution, has been
used in this study. Besides Satellite variables LSTd, LSTn,
NDVI and NDBI, other six geographical and topographic
variables are considered: latitude, longitude, distance to
coast, altitude, orientation and slope of the terrain for every
meteorological station and pixel. Furthermore, the calendar
day (cd) has been transformed into a new calendar day (cd*)
according to a cosinus transformation (equation 1). These
variables are used to fit an OLS model, with the air
temperatures obtained from the meteorological stations as
the dependent variable. This model makes it possible to
know (at a resolution of 1 km / pixel) the spatial distribution
of the air temperature.
2𝜋(𝑐𝑑 − 𝑐𝑑𝑚𝑎𝑥 )
𝑐𝑑 ∗ = cos (1)
365
6. Once the air temperature model developed from the
meteorological stations has been obtained, said model can
be merged with the information resulting from the high-
resolution satellite sensors (Landsat 8). In this way, the UHI
derived from the air temperature can be visualized in much
more detail. Therefore, to downscale the multiregression
model developed in section 5, the variables (NDVI, NDBI,
LSTn, albedo, ...) are used at a resolution of 30 m/p
(obtained from Landsat).
7. The study of the spatial distribution of the Urban Heat Island
of the Barcelona Metropolitan Area in August 2015 will
allow determining the health risk posed by high
temperatures derived from heat waves.
8. Determine the green areas of the MAB through the use of
remote sensing (Sentinel 2), as well as study the degree of
urban greenery of the different landscapes. Sentinel 2 offers
a higher spatial resolution (10 meters/pixel) than Landsat 8
(30 m/p), which helps to more accurately delimit green
spaces. NDVI has been used as an indicator of vegetation
quality. Spaces with a surface area greater than 1,000 m2 and
an NDVI greater than or equal to 0.3 have been considered
"urban parks".
9. And finally, model the effect of greenery in the nocturnal
UHI as well as quantify the effect that the increase in green
 spaces (measured by greenery characteristics and distance
to parks) would have on temperature regulation.
3. RESULTS
3.1 Heatwaves and Mortality
Figure 2 shows the (official) weather stations existing in the
Barcelona Metropolitan Area. Among them stands out, due to its
high temperatures, the Raval station.
Figure 2. Weather stations
The European Union’s Copernicus program highlights two
climatic indicators: summer days, when the temperatures exceed
25 degrees, and tropical nights, when the minimum temperature
does not drop below 20 degrees. For the period of 1971 to 2020,
the Raval was the weather station with the highest number of
summer days per year: 114.04. Almost four months/year were
summer days. Regarding tropical nights, the Raval on average
between 1971 and 2020 reached 66.94 nights or over 2 months a
year.
The application of the criteria for defining heatwaves in
Barcelona, which is at least 3 consecutive days with a maximum
temperature (DHW) or a minimum temperature (NHW) higher
than the 95th percentile for the months of July and August in the
period 1971 to 2000, gave us the thresholds of 32.49ºC by day
and 23.86ºC by night for Raval station (located in the old city).
Applying these thresholds, 18 daytime heatwaves would have
been produced in the center of Barcelona, with 76 extremely hot
days between 1971 and 2020. The Raval station would have
endured 70 nighttime heatwaves and 577 very hot nights (table
1).
Years NHW HN
1971-1980 0 3
1981-1990 4 28
1991-2000 8 66
2001-2010 24 192
2011-2020 34 288
Total 70 577
Table 1. Nighttime Heat Waves and Hot Nights (Raval)
As can be seen, NHW and extremely hot nights (HN) increased
almost exponentially, the first from 0 (1971 to 1980) to 34 (2011
to 2020), and the second from 3 (1971 to 1980) to 288 (2011 to
2020). If the growth continues at the current rate, it could lead to
140 extremely hot nights per year in the center of Barcelona in
2050.
As we have previously indicated, it is during the night when
excess temperature can generate more negative effects on health,
causing an increase in premature deaths. Using the daily
mortality data for Barcelona provided by the Barcelona Public
Health Agency (ASPB), from January 1 to December 31, 2018,
this study seeks to assess the effects of high temperatures on
premature mortality. Figure 3 shows the daily deaths in the city
of Barcelona in relation to the minimum temperatures (Tm). The
minimum temperature threshold that marks the minimum deaths
is located at 21.3ºC.
Figure 3. Deaths vs. Minimum Temperatures (2007-2018)
A total of 852 days between 2007 and 2018 experienced night
temperatures above 21.3 degrees, which represented an increase
in mortality that can be estimated at 995 additional deaths in the
same period. Between 20 and 23.86 degrees Celsius (the limit at
which a nocturnal heatwave is identified), the increase in
mortality is very low (1 additional death). However, above this
limit, mortality multiplies exponentially: at 25 degrees (the
threshold that we call a torrid night) there are 2.81 additional
deaths, which increases to 5.83, 10.98, 19.34, 32.27 and 51.57 for
each additional degree. The warming forecasts for the Barcelona
area, established in the previous subsection and in the event of no
human adaptation to the climate (and the massive installation of
air conditioning systems in homes), may lead to a very marked
increase in premature mortality due to night heat.
3.2 Nighttime LST
To proceed with the analysis of daytime and nighttime LST,
Landsat 8 images relative to the month of August 2015 were used
(8/3, 8/28). To obtain a completely cloud-free image of the study
area, the night thermal images of 8/28 and 9/13 were combined.
Figure 4 compares the August 2015 AMB LST day and night
images from MODIS and Landsat.
Figure 4. MODIS vs. Landsat LST (day and night)

As is well known, the low spatial resolution of MODIS contrasts
with the better resolution of Landsat. However, the few nighttime
thermal images from Landsat have made detailed knowledge of
nighttime LST difficult. This study focused on heatwave events
occurred in August 2015, for which there are nighttime Landsat
images, which make it possible to study the LST-UHI at night.
Figure 4 shows the big difference between UHI during day and
night. The UHI appears very clearly, especially when compared
to the daytime LST (on the left in the figure). While during the
day the UHI does not stand out, at night the greater thermal inertia
of the urban materials, buildings and roofs determines the greater
heat gradient of the built-up area compared to the rural ones. The
metropolitan mean of the LST reaches, at night, 22.6°C,
confirming that we are faced a Night Heat Wave event. Nighttime
UHI is very pronounced: 2.9 degrees higher between rural
(21.23°C) and urban (24.10°C) land covers, means a 13.5%
higher. It can be affirmed that a significant part of urbanized
landscapes had a temperature typical of a “torrid night” in the
date of study (August, 2015).
Figure 5 presents the result of modelling nighttime temperature.
This model explains 90.9% of the variation in nighttime land
surface temperatures, with all significant variables. Table 2
shows this “hybrid” model. Among the explanatory variables,
winter NDVI and MODIS LST stand out. These two variables
alone explained 88.1% of the spatial variation of Landsat LST.
The inclusion in Step 3 of the variable artificialization raised the
R2 to 0.892. However, in Step 4 the variable distance to the coast
was incorporated into the model, with the sign changed due to its
interaction with artificialization. These four variables accounted
for over 90% of the explanatory power of the OLS model.
Figure 5. Nighttime LST-UHI model
Table 2. “Hybrid” OLS model oh nighttime LST
3.3 Modelling Land Surface Air Temperature
Once the nocturnal LST has been obtained, the air temperature is
modeled. A multi-regression model has been developed with
night air temperature (obtained by weather stations) as dependent
variable, and nighttime and daytime LST (obtained by MODIS),
latitude, longitude, altitude, slope, orientation, distance to the sea,
calendar day* (cd200), NDVI and NDBI as independent
variables. The model has been developed throughout the
extended Metropolitan Area of Barcelona for all non-cloudy days
of the year. The MAB only has 13 weather stations (see figure 2).
We have used the 20 stations closest to the city center to improve
the significance of the regression model.
The model, with 1832 observations, reaches an R2 of 0.943, with
7 significant variables. The most significant variable is the
MODIS nighttime LST, followed by distance to the sea (-), slope
(+), longitude (-), NDVI (-), NDBI (+) and calendar day*
(cd200). Altitude and latitude are not incorporated into the best
model due to their high collinearity with the variable distance to
the sea. The NDVI and the NDBI with the correct sign. In step 7
the variable calendar day* (cd200) is incorporated; its relatively
low significance (student's t = 6.579) is due to its high collinearity
with the MODIS LST (Pearson's coefficient = 0.911). Table n. 3
presents the regression coefficients of the model.
Table 3. Nighttime Air Temperature Model (MODIS, 2015)
Figure 6 shows the air temperature model obtained by MODIS,
merging it with Landsat 8 information (urbanized land, line blue).
This fusion allows to see with great detail the LSAT-UHI of the
Metropolitan Area of Barcelona (with a resolution of 30 meters /
pixel).
Figure 6. Nighttime LSAT-UHI model
The metropolitan mean of the LSAT reaches, at night, 20.00
degrees Celsius, with 18.34°C in rural covers and 21.89°C in
urban ones, 19.4% higher.
Figure 7 shows the metropolitan air temperature of Barcelona
(August 2015):
 In white, areas with an air temperature below 20 degrees in time were used. Figure 8 shows these “parks” in the city of
Celsius. This area basically coincides with rural, Barcelona.
nonurbanized areas.
More than 80% of the inhabitants who live in the central city of
 Also below 20°C some of the main parks and green areas of Barcelona are more than 200 meters from a true urban green area.
Barcelona, such as Montjuich, the Citadel or the Tres In fact, the standard of green spaces per inhabitant does not reach
Turons. 5 m2 per inhabitant (if Collserola park, located outside the urban
 In pink, the areas with an air temperature between 20 and 22 continuum, is excluded). Only in areas of lower density, of urban
degrees, characterized therefore by having a relatively cool sprawl, in the metropolitan periphery is the proximity to greenery
"tropical" night. This degree of heat makes it difficult to rest less than 200 meters.
at night, but it has no relevant effects on health. None of
these areas is located in the urban continuum of Barcelona,
characterized by its high density. Only the Urban Sprawl
areas, located on the northern slope of the Collserola
mountain chain, reach this temperature.
 In red, the areas with an air temperature between 22 and 24-
25 degrees Celsius. This degree of heat does not only imply
discomfort. It also has a significant effect on health,
representing the threshold above which premature deaths
due to heat increase significantly. These areas correspond to
the urban continuum of Barcelona. Neighborhoods
characterized by their high density.
 Finally, in dark brown, the areas with a temperature above
24-25 degrees. Temperature that represents serious
problems for human health ("torrid" nights) if there is no air
conditioning installation in the homes. These areas coincide
with the urban center of the city ("Ensanche"), as well as,
above all, in the historic center of Barcelona ("Ciutat Figure 8. Green areas (>1000 m2)
Vella").
In addition to the NDVI, or others vegetation indexes (like Soil
Adjusted Vegetation Index - SAVI, Enhanced Vegetation Index
- EVI, Leaf Area Index - LAI, FAPAR, Fcover, …), proximity to
parks has a relevant effect on Urban Heat Island. Usually, the
parks’ cooling effect is calculated by its extent and intensity. The
cooling intensity (ΔT) is the difference between the temperature
of the parks (PT) and the surrounding urban areas (SUT),
calculated as SUT - PT. The cooling effect shows positive ΔT
values. The cooling extent limit (Lmax) is the maximum distance
reached by the microclimatic influence of the park. A positive
cooling effect implies lower temperatures near to the park and a
gradual increment with increasing distance. In a fitted dataset of
temperature of urban spaces ordered by distance to a park, the
cooling effect generates a “cooling curve”, which ends at the
maximum ∆T (∆Tmax) and defines the Lmax point. Although
this methodology is usually applied at local scale, showing the
Figure 7. Nighttime LSAT and Risk extent and intensity of the cooling generated by each urban green
area, in this paper a statistical approximation was made for the
3.4 Urban greenery and Urban Heat Island entire metropolitan area, thus obtaining the average extent of park
cooling of the entire the metropolis.
As mentioned above, the increase and improvement in vegetation
measured by the NDVI among other indicators is a fundamental
element of urban cooling. In this work, several models have been
developed to evaluate the effect of urban vegetation on the night
air temperature of the urbanized land of Barcelona Metropolitan
Area. The model with only the NDVI as the explanatory variable
explained 60.5% of the spatial variation of LSAT, which implied
that an increase of 0.1 in the NDVI represented nighttime cooling
of 0.7 degrees Celsius.

In addition to the NDVI, proximity to parks had a relevant effect

on urban heat. However, to determine to what extent vegetation
enhancement actually affects urban climate, it is necessary to
determine what is really green in terms of impact on human
Figure 8. Temperature vs. Distance to Green Areas
health and well-being. In this study we considered parks to be
green areas (public or private) with a winter NDVI greater than
Figure 8 shows the land surface temperature in relation to the
0.3, and an area > = 1,000 m2. To better resolve what was “really
distance to the green areas of the urbanized land in the Barcelona
green”, the Sentinel 2 images (10 meters/pixel) that were closest
Metropolitan Area. The adjustment by means of non-linear
regression (NLR) of a fourth-degree polynomial model, helps to
understand the cooling effect generated by urban greenery.
By itself, the distance to urban greenery explains 43.6% (R2) of
the spatial variation in temperature. The adjustment of the model
allowed the deduction of an average extension of cooling of the
parks of 178 m. Given that in the city of Barcelona most of the
inhabitants live more than 200 meters from a park, the beneficial
climatic effect of green areas on the population is limited.
The adjustment of an explanatory linear model of the
temperatures, with only the distance to the green area as an
independent variable, showed that for every 1,000 away from the
parks, there was a cooling of 5.7 degrees. However, as indicated,
the cooling effect of parks is not linear. In addition, there is
collinearity between the distance and the NDVI, so an OLS
model is developed with the NDVI and the distance to greenery,
to evaluate more precisely the beneficial effect of proximity to
greenery.
The adjustment of an explanatory linear model of the
temperatures, with NDVI and distance to greenery, allows
evaluation of the beneficial effect of vegetation and proximity to
greenery. This joint model explains 61.6% of the spatial variation
of the LST, suggesting that an increase in 0.1 points in NDVI
would represent a reduction in temperatures by - 0.645ºC. Table
4 shows this model. A distance of up to 1,000 from the parks
would imply a temperature increase of 1.438ºC. Therefore, urban
policies can be developed that reduce the risks of excess heat on
human health. Urban greening, together with land perviousness,
is especially relevant to prevent high temperatures.
Table 4. Nighttime Air Temperature Model (MODIS, 2015)
Therefore, the models suggest that an improvement in urban
vegetation that represents a significant increase in NDVI and a
significant reduction in the distance from homes to parks would
have very beneficial effects on health and well-being. Night
cooling would undoubtedly reduce the risk of premature death in
heatwave events. Especially in an area as dense and sparse in
vegetation as the city of Barcelona.
4. CONCLUSIONS
In this paper, we analyzed the significant role of the urban heat
island due to the increase of global warming in the spatial field
of study, Barcelona Metropolitan Area. We also verified the
extent of warming since the middle of the last century. An
increase in heatwaves and extreme temperatures was detected.
We tried to demonstrate the effect of this progressive warming
on health, and especially on mortality, since the rise in
temperatures in the warm months of July and August has the
main effect of increasing premature mortality, in addition to
aggravating various other health effects.
The starting hypothesis of the paper was that the main adverse
health effects occur at night, due to the difficulty of adapting the
body (in the context of a built-up area without widespread access
to air conditioning) to the night heat. In cases of high nighttime
temperatures, heat stress persists and is aggravated by the fact
that the human body cannot rest at night. The most common
impact of hot nights on human health is on sleep and rest. Heat,
along with other factors, can cause sleep disturbances and
deprivation due to the necessary thermoregulation processes. In
the context of climate change, in Mediterranean towns there has
been a notable increase in the number of tropical and torrid
nights.
Given the relevance of temperatures during the night, this study
focused on the study of extreme heat phenomena that occurred in
Barcelona Metropolitan Area, defining the scope of heatwaves
and extreme temperatures at night. In this context, UHI increase
the effects of heatwaves, and represent a serious inconvenience
to human health and comfort. Consequently, the spatial
distribution of LST at night was studied to define the concept of
night UHI, using the information from various space sensors, and
especially from Landsat 8. At night and during the warm months
of the year, there is a high correlation between the LST and the
air temperature supported by humans (LSAT). Therefore,
knowledge of the LST is of vital importance due to its decisive
role in the climate process and as a real indicator of air
temperature, which is what really affects human health and
comfort.
For this reason, a model of the spatial distribution of the LSAT
was also developed using the information provided by MODIS.
Said LSAT model confirmed the main role played by the LST in
explaining nighttime air temperatures, in addition to allowing
downscale the model to 30 m/p. Change of resolution of the
LSAT model that allowed better visualization of the UHI. And
also to evaluate the health risk of nocturnal heat waves in the
different urban landscapes of the Barcelona metropolis.
In this context, the research has aimed to demonstrate the
important role played by urban vegetation in adaptation to Global
Warming and Urban Heat Island. Green areas have lower
temperatures than other land uses and generate a cooling effect
that extends to its surroundings creating a "cool island" effect.
The second hypothesis, therefore, has been aimed at
demonstrating said cooling role in a heat wave context in the case
study, the Barcelona Metropolitan Area. The pronounced scarcity
of vegetation, as well as the high nighttime temperatures (often
above the comfort threshold) make the city of Barcelona a place
that is not very resilient to Global Warming, as well as to Heat
Waves.
The development of an explanatory model of the role played by
vegetation in the UHI makes it possible to determine the effect of
urban green on temperatures, and therefore to design urban
policies aimed at increasing adaptation to climate change and
thus reducing negative impacts on health and the comfort. The
adjustment of the model allowed the deduction of an average
extension of cooling of the parks of 178 m. Given that in the city
of Barcelona most of the inhabitants live more than 200 meters
from a park, the beneficial climatic effect of green areas on the
population is limited.
The lack of green areas in the city of Barcelona is very
pronounced. In fact, the standard of green spaces per inhabitant
does not reach 5 m2 per inhabitant. This fact demonstrates the low
resilience of the city in the face of global warming and the
progressive increase in heat waves. Therefore, the models
suggest that an improvement in urban greenery that represents a
significant increase in NDVI and a significant reduction in the
distance from homes to parks would have very beneficial effects
on health and well-being. Night cooling would undoubtedly
reduce the risk of premature death in heatwave events. Especially
in an area as dense and sparse in vegetation as the city of
Barcelona.
 ACKNOWLEDGEMENTS Parsons, K., 2014. Human thermal environments. The effects of
hot, moderate and cold environments on human health, comfort
The study is part of the project "Extreme Spatial and Urban and performance. 3rd edn. CRC Press, 1-635.
Planning Tool for Episodes of Heat Waves and Flash Floods. Royé, D. 2017. The effects of hot nights on mortality in
Building resilience for cities and regions", supported by the Barcelona, Spain. International Journal of Biometeorology, 61,
Ministry of Science and Innovation of Spain. The authors also 2127–2140.
thank the Barcelona Public Health Agency for the information
provided on daily mortality in the city. Serra, C., Lana, X., Martínez, M.D., Roca, J. Arellano. B., Biere,
R., Moix, M. & Burgueño, A. 2020. Air temperature in Barcelona
metropolitan region from MODIS satellite and GIS data,
Theoretical and Applied Climatology, 139: 473–492.
REFERENCES Sobrino, J.A., Jiménez-Muñoz, J.C., Sòria, G., et al. 2008. Land
surface emissivity retrieval from different VNIR and TIR
Arellano, B. & Roca, J. 2021. The International Archives of the sensors. IEEE Trans. Geosci. Remote Sens, 46, 316–327.
Photogrammetry, Remote Sensing and Spatial Information Spronken-Smith, R. & Oke, T., 1998. The thermal regime of
Sciences, Volume XLIII-B3-2021, 15-22. urban parks in two cities with different summer climates.
Arellano, B. et al, 2021. Modelling nighttime air temperature International Journal of Remote Sensing, XIX(11), 2085-2104.
from remote sensing imagery and GIS data. Proc. SPIE 11888, Tan, J., Zheng, Y., Tang, X. et al. 2010. The urban heat island
Space, Satellites, and Sustainability II, 118880H. and its impact on heat waves and human health in Shanghai. Int
Benali, A., Carvalho, A.C., Nunes, J.P., Carvalhais, N. & Santos, J Biometeorol 54, 75–84.
A., 2012. Estimating air surface temperature in Portugal using
MODIS LST data. Remote Sensing and Environment, 124, 108–
121.
Bowler, D., Buyung-Ali, L., Knight, T. & Pullin, A. 2010. Urban
greening to cool towns and cities: A systematic review of the
empirical evidence. Landscape and Urban Planning, 97-3, 147-
155.
Eurostat, 2022. Statistics Explained.
https://ec.europa.eu/eurostat/statistics-
explained/index.php?title=Population_structure_and_ageing
(accessed on 30-03-2022).
Gasparrini, A. et al, 2015. Mortality risk attributable to high and
low ambient temperature: a multicountry observational study.
Lancet, 386, 369–75.
Guzinski, R., Nieto, H., Sandholt, I. & Karamitilios, G., 2020.
Modelling High-Resolution Actual Evapotranspiration through
Sentinel-2 and Sentinel-3 Data Fusion. Remote Sens. 12, 1433.
IPCC, 2012. Managing the Risks of Extreme Events and
Disasters to Advance Climate Change Adaptation. Special
Report of the Intergovernmental Panel on Climate Change.
Cambridge University Press, 1-582.
IPCC, 2021. Climate Change 2021. The Physical Science Basis.
Working Group I contribution to the Sixth Assessment Report of
the Intergovernmental Panel on Climate Change. Cambridge
University Press, 1-3949.
Kuttler, W., 2012. Climate Change on the Urban Scale – Effects
and Counter-Measures in Central Europe. In: N. CHhetri, ed.
Human and Social Dimensions of Climate Change. Rijeka,
InTech, 105-142.
Mahour, M., Tolpekin, V., Stein, A. & Sharifi, A., 2017. A
comparison of two downscaling procedures to increase the spatial
resolution of mapping actual evapotranspiration. ISPRS Journal
of Photogrammetry and Remote Sensing, 126, 56-67.
Meehl, G.A. and Tebaldi, C., 2004. More Intense, more Frequent,
and longer lasting Heatwaves in the 21st Century, Science, Vol.
305, Issue 5686, 994-997.
Mendelsohn, R., Kurukulasuriya, P., Basist, A., Kogan, F. and
Williams, C., 2007. Climate analysis with satellite versus
weather station data. Climatic Change, 81, 71–83.
Oke, T.R., 1967. City size and the urban heat island. Atmospheric
Environment, Vol. 7, Issue 8, 769-779.